IC/86A8 IHTRODUCTIOH INTERNAL REPORT 1. Development of solar energy research programmes must start with a study of solar energy available at the site or region of interest. Long term measurements of solar radiation, consisting mainly of solar radiation on a International Atomic Energy Agency horizontal surface, however, exist for relatively few meteorological stations. and For places where it is not directly measured, solar radiation can be estimated ; United Nations Educational Scientific and Cultural Organization by interpolation from nearby localities where radiation data are available by INTERNATIONAL CENTRE FOR THEORETICAL PHYSICS using models and empirical correlations.

One such method is the correlation found by Angstrom [l]. Black et al. [2] and others between global radiation and the duration of sunshine, which is ESTIMATING SOLAR RADIATION IN GHAHA • measured at many meteorological stations. This correlation has been used by many authors (e.g. Lof et al. [3]) to draw solar radiation maps with better details than would be possible using only directly measured radition data.

K. Anane-Fenin ** The present work stems from the need of knowledge of solar radiation data International Centre for Theoretical Physics, Trieste, Italy. in and to fill the gap of the radiation data for the places lacking direct radiation measurements. In this paper the available data on global solar radiation in relation to recordings of sunshine duration for Ghana are examined.

ABSTRACT 2. GLOBAL SOLAR RADIATION AND SUNSHINE HOURS CORRELATION The estimates of global radiation on a horizontal surface for 9 towns in Ghana, West Africa are deduced from their sunshine data using two methods Global solar radiation is measured at 16 meteorological stations, whilst developed by Angstrom and Sabbagh et al. An appropriate regional parameter is 2l( stations record sunshine duration in Ghana. No station measures diffuse solar determined with the first method and used to predict solar irradiation in all radiation. Solar radiation (global) is measured with pyranometer of the Bellami the 9 stations vith an accuracy better than 15%. Estimation of diffuse solar type, and sunshine duration with Campbell-Stokes tropical sunshine recorder. irradiation by Page, Lin and Jordan and three other authors' correlation are The data used in this paper are monthly averages for global solar radiation on a 1 performed and the results examined. horizontal surface from 1972-197 *, and sunshine duration hours averages over many years. These were supplied by Ghana meteorological services in .

Of the many models, the most popular is the regression equation of MIRAMARE - TRIESTE the Angstrom [l] April 1986 (1)

The above equation is used to determine a and b for nine Ghanian stations, which are videly distributed. They are Accra, Takeradi and Saltpond along the southern coastal belt, Kumasi, Ho and in the central forest region, * To Ije submitted for publication. and Tamale, Bole and Ttendi in the northern savannah region. The geographical •" Permanent address: Physics Department, The University of Cape Coast, Cape Coast, Ghana. locations of the nine towns are shown in Table 1.

-2- Sabtagh et al. [1+] used the average hours per month of sunshine n in their equation ~ = 0.96k - 0.786 n/a (9) H H •> ii(a + b In m) (2) StanhillllO]. where m = l,2,3,1t,5>6t6,5t1*>3»2,l. The correlation is used on data from three stations. — ' 0.775 + 0.00653 (HI - 90) - [0.505 + 0.0 + 1*55 (a - 90)] COB[115K_, - 1031 (10) i) Method A - The mean monthly daily global solar radiations from each of the nine Collares-Pereira and Rabl [11]. stations are calculated from Eq. (l) (see Table k). In these calculations the values of H are calculated according to [5] k. RESULTS AND DISCUSSION ^gs- D I] [cos 4 cos 6 sin o>a + rg^r- sin $ sin fi] The variation of monthly mean daily solar radiation for the nine towns (3) where are shown in Figs. 1-3. They are similar in pattern. Peak insolation occurs cos u = -tan $ tan 8 in April-May and then in October-November. Yendi which is in northern Ghana 3 _2 has an annual average of daily irradiation of 22.0 MJm . The lowest annual The values of N, the average day length for a given month is computed from _2 average Of 13.0 MTm occurs in Kumasi in July. — 2 —1 N = — cos (-tan $ tan 6 )• (5) The predicted average values of H using Eq. (l), the regression The least square method is used to calculate the coefficients a and b for coefficients a, b and their correlation coefficients for each station are different locations, presented in Tables 2 and I*. It is seen that the correlation is good for all towns except for Ho,where r = 0.82. In general,the agreement is fair with ii} Method B - This method requires only hours of sunshine as the input the estimated values of Eq. (1). It shows a better accuracy for the summer parameter. The results of the computed solar radiation for the three towns are months than for the winter months. For example with the exception of Kumasi and shown in Table 6. Takoradi where the discrepancy reaches 15$ and 19% in the respective months of September and August, the difference throughout the year for the nine towns during 3ummer is less than 10/S. The predicted values show an excellent agreement 3. PREDICTION OF DIFFUSE RADIATION with the measured values for the towna of Wenchi, Yendi and Bole, throughout As there is no information on diffuse solar radiation at any station in the year, the difference being less than ± 12?. Ghana, we have to resort to theoretical methods for its evaluation. A well itnown Prediction values by the regional parameter (Table 3) also fits the relation for this purpose is the correlation equation developed by Page [6] measured values well as shown in Table 5. Table 3 shovs that the sum a + b which represents the clear day fraction of H increases both southwards and TA = 1.00 - 1.13 K_, (6) northwards from the . The annual average of n/N also shovs H similar variations. These trends are concurrent with the general observation of Another commonly used correlation due to Liu and Jordan [7] and developed by the decrease in rainfall as one moves away from the central region to the north Klein [8] or south. ii. _ _2 _r, J^ " 1.39 - ^.027 (L, + 5.531 \ - 3.108 K£ . (7) The monthly average daily solar irradiation data computed by Eq. (2) for : H Three other methods are used: the three towns are presented in Table 6 where a comparison with measured values is also given. The predicted values 3how better accuracy in winter than in ^ = 1 - n/N (8) summer, where the discrepancy reaches as high as S6% for Kumasi and Takoradi. H [9]. -It- -3- The seasonal variations of global radiation for the three regions are shown i) The pattern of monthly and seasonal fluctuations of global solar irradiance in Figs. h-S, In the which is.mostly dry, radiation is fairly are almost the same for all the nine towns. Peak global irradiance occurs in high throughout the winter months, reaching 680 MJm at Yendi in May. Other April-May and another lower peak in October-November. regions have lower irradiance during the rainy season (June-September). The ii) The annual global solar radiation of these towns rangeB between 5700 and central region seems to be characterized by lower irradiance through the year. 71*00 MJm . Hence the abundance of solar radiation in the populated areas of In general the annual total global solar radiation for the nine towns lies Ghana is established. between 5700 and 71'00 MJa~S. iii) It is evident from Table 8 that the monthly average value of the ratio H /H for Tamale is quite low, suggesting a low atmospheric turbidity for the

5. COMPUTATION OF DIFFUSE RADIATION Northern region of Ghana.

The values of monthly variation of diffuse solar radiation shown in Figs. 7-9 have been computed using Eqs.QS'-lO1). It is seen that the variation ACKNOWLEDGMENTS follows the same pattern throughout the year. The predicted values obtained The author would like to thank Professor Abdus Salsa, the International by Eqs. (8) and (9) however are much higher. The minimum H. appears in November Atomic Energy Agency and UNESCO for hospitality at the International Centre for d Theoretical Physics, Trieste. He would also like to thank Professors L. Bertocchi for Tajtiale whilst the maximum values appear in August-September. The maximum and G. Furlan of the ICTP, Trieste,and Dr. F. Stravisi of CNR - Istituto values for Accra and Kumasi appear in March-April and August-October. In the Talassografico di Trieste for their deep interest throughout the work and for absence of experimental data it is difficult to Judge which method is better. going through the manuscript. However, it should be noted that the average yearly values of the ratio

Hd/3 for the three towns lie between O.3l4 and 0.59. The ratio, H /fi which is the variation due to atmospheric conditions (dust, smoke, water vapour and suspended matter) has been given a number of values such as 0.12 - 0.25 [7] 0.30 - 0.36 [10] and 0.30 for tropical locations [12], The consensus value seems to be about 0.30 which makes Eq. (7) the most appropriate for the estimation of diffuse radiation in Qhana.

6. CONCLUSION

It is possible to compute global solar radiation on a horizontal surface at any location in Ghana using Eg.. (1). A comparison of the two methods used shows that Method A gives better results than Method B. It is therefore possible to determine clear-day global solar irradiance at any location in the country using the appropriate regional coefficients in Table 3.

Table 5 indicates that apart from some few winter months (where error is about ± 2k%) the error in the computed values using regional parameters is expected to be in the range of ± 1%.

The diffuse solar irradiance for the country can be estimated by means of EG> (7). However the correlations of Eqs. (6)-(10) require further experimental verifications, Furthermorejthe results suggest the following concluding points: -6- -5- w t

REFERENCES TABLE 1

Geographical locations of towns used in the study [l] A. Angstrom, "Solar and terrestrial radiation", Q.J.R. Meteorol. Soc.

5p_ 121 (1921*). Station Latitude Degrees [2] J.R. Black, C.W. Bonython and J.A. Prescott, "Solar radiation and Accra duration of sunshine", Q.J.R. Meteorol. Soc. 60, 231 (1951*). 5.33 N Takoradi 1*.59 M [3] G.O.G. Lof, J.A. Duffle and CO. Smith, "World distribution of solar Saltpond 5.12 N radiation", Rep. Ho. 21, Engineering experiment station Madison 66 (1966). Kumasi 6.1(1 H [1*] J.A. Sabbagh, A.A.M. Sayigh and E.M.A. El-Salam, "Estimation of the Ho 6.35 N correlation of solar radiation and sunshine duration in Riyadh, Saudi Arabia", Wenchi 7.1*2 H Pak. J. Sci. Res. l£, 6 (1973). Tamale 9.25 N

[5] J.A. Duffle and W.A. Beckman, Solar Energy Thermal Processes (Wiley Interscience, Yendi 9.26 N New York, I960). Bole 9.02 N

[6] J.K. Page, "The estimation of monthly mean values of daily total short-wave TABLE 2 radiation on vertical surfaces from sunshine records for latitude !*0°N - !*0°S", Proc. U.K. Conf. on New Sources of Energy, Paper No. S/98 (1961). Regression and correlation coefficients

[7] B.Y. Liu and R.C. Jordan, "The interrelationship and characteristic Station n/N(annual av.) a b a + b r distribution of direct diffuse and total solar radiation", Sol. Energy 1*_(3), Accra 1 (I960). 0.51* 0.29 0.1*7 0.76 0.90 Takoradi 0.50 0.25 0.1*7 0.72 0.93 [8] E.A. Klein, "Calculation of monthly average insolation on tilted surfaces", Saltpond O.56 0.26 0.1*5 0.71 0.88 Sol. Energy 19.(1*), 325 U97T). Kumasi o.Vr 0.25 o.Ul* O.69 0.93 [9] M. Iqbal, "Correlation of average diffuse and beam radiation with hours of Ho 0.51* 0.21 0.1*6 0.6a 0.82 bright sunshine", Sol. Energy 23., 169 (1979). Wenchi O.58 0.28 0.36 0.71* O.89

[10] G. Stanhill, "Diffuse sky and cloud radiation in Israel", Sol. Energy 1£, Tamale 0.63 0.27 0.U7 0.1k 0.9l» Yendi 96 (1966). 0.61* 0.32 0.1*1 0.73 0.89 Bole 0.60 0.28 o.Wt 0.72 0.92 [11] M, Collares-Pereira and A, Rabl, "The average distribution of solar radiation correlations between diffuse and hemispherical and between daily and hourly TABLE 3 insolation values", Sol. Energy £2(2), 155 (1979). The regional parameters [12] A.J. Drummond, "On the measurement of sky radiation", Arch. Meteorol. Vienna, Region n/H a b a + b BX, 139 (1956). Coastal south 0.533 0.267 0.1*63 0.730 [13] H. Iqbal, An Introduction to Solar Radiation, (Academic Press, Toronto, 1983). Central region 0.530 0.2l*7 0.1*20 0.667 northern region 0,623 0.290 o.Uo 0.730

-7- TABLE 1* - Comparison of measured and estimated values of H by Eg., (l)

STATION JAN FBB MAE APR MAY JUN JUL AUG SEP OCT N0V DEC Accra H 16.77 19.23 20.52 21.31 20.86 15.94 16.92 17.17 18.65 20.93 20.32 16.72 H 19.26 21. £1.40 21. 16.24 16. 17.41 02 20. c 07 07 19.92 87 19. 71 20.92 18.36 % Error -ll*.97 -9.57 -4.29 1.13 4.51 -1. 86 0.30 -l.4o -1. 99 1.05 2.95 -9.81 Ta&oradi H 15.92 18.57 20.15 20.41* 18.22 14.21 15.1*0 15.62 17.61 18.66 20.09 16.20 17.62 26 16.85 14.61 22 12.65 18.01 Hc 19- 19.79 19.51* 15. 15.99 19.40 18.00 % Error -10.68 -3.72 1.79 1*.40 7-52 -2.81 1.17 19.01 9.20 3.48 3.43 11.11 Saltpond H 15.61 17.94 19.61 19.70 18.43 11*.43 14.63 15.62 16.38 17.41 20 11 20.02 H 18.22 £0.06 20.01 19.67 18.24 14 63 15.37 16.50 17.1.0 19 70 19 78 18.53 % Error -16.73 -U. 82 -2.00 0.15 1.03 -1. 39 -5.06 -5.63 -6.23 -13 15 1.60 7.44 Kumasi H 15.11* 16.76 17.92 18.47 18.31 15 68 13.38 12.71 14.12 15 57 17 64 14.29 15.76 70 18.66 11 Hc 17 18.53 IT 70 15 13.16 13.66 16 19 15 81 16 70 15.52 % Error -4.23 -5 61 -3 40 -1. 03 3 33 3 64 1.6U 7.47 -14 66 1 54 5 33 8.61 Ho H 14.50 16 Uo 17 16 18.42 17 92 15 50 14,55 13.59 14 41* 17 73 18 1*5 14.20 H 16.46 18 00 18.15 17 93 17 47 14 62 12 74 12.71* 14 36 17 33 18 06 17.26 % Error -13.66 -9 76 -5 77 2 66 2 51 5 68 12 44 6.25 0 55 2 26 2 11 -21.55 Wenchi H 15.77 17 19 2b 19 30 19 32 16 B7 15 30 14.39 14 32 16.& 17 63 14.93 i 16.26 18 64 18 82 18 93 18 39 16.1*7 14 87 15.06 15 75 17.69 18 39 16.53 % Error -7.67 -1*.02 2 26 1 92 4.ttl 2 .37 2 81 -4.66 -9.99 -6.31 -1* 31 -10.72 Tamale Jj 19.71 20 .51+ 18 01 20 65 20 .29 20 29 18 96 17.43 15.69 20 83 19 .74 16.27 H 20.01 21 .12 21 21 21 .64 21 .08 c 29 39 19.75 17 58 15.92 17.69 21 .12 19.33 % Error -1.52 -2 .82 -19.ob1 -3.58 -6.65 2.66 7.38 8.66 12 .75 -1 .20 -6.99 18.81 Yendi H 19.50 21 .00 21 .64 22.02 22.07 20.80 19 .01 17.23 18.1*9 21.1*8 20.98 17.99 H 20.20 21.1*8 22 23 21.83 S2.00 20.00 17 • 97 IT. 48 18.68 21.34 21.43 20.00 % Error -3.59 2.29 -2.73 0.66 0.32 3.35 5.V7 -1.45 -1 .03 0.65 -2.14 -11.10 Bole H 19.59 21.63 20.70 21.43 20.88 19.53 17.1*3 15.98 17• 19 £0.20 20.111; 17.27 H 20.14 20.91 £0.91 20.71 20.74 18.79 16.1*5 15.57 15.78 19-96 19 .86 18.74 % Error -2.81 3.33 -1.01 3.36 0.67 3.79 5.62 2.57 8.20 1.20 2.84 8.51

TABLE 5 - Comparison of measured and estimated values of H using regional parameters

KEGIOH JAN FEB MAR APE MAY JUH JUL AUG SEP OCT NOV DEC Coastal (south) Accra H 18.40 20.13 20.43 20.11 18.99 15.1*5 16.01 16. 54 18.11 19.77 20.01 17.51 % Error -9.72 -4.68 0.44 5.63 8.96 3.07 5.38 3.67 2.90 5.54 1.53 -4.72 Taioradi H 18.11 19.76 20.31 20.06 17.41 15.16 15.79 13.30 16. 59 16. 53 19.86 18.46 % Error -13.16 -6.Uo -0.79 1.86 4.1*5 6.83 -2.53 14.85 5.79 0.70 1 ll* -13.95 Saltpond H 18.77 20.66 20.62 20.27 18.80 15.10 15.86 17.02 17.91* 20.29 20 37 19.08 % Error -20.24 -15 16 -5.15 -2.89 -2.01 -4.64 -8.1*1 -8.96 -9.52 -l6. 54 -1 29 4.70

Central Kumasi H^ 15.41* 17 30 18.10 18.21* 17.30 lit.83 13.00 13.46 15.87 15. 50 16 32 15.18 % Error -1.98 -3.22 -1.00 1.25 5-52 5 1*2 2 64 5 90 -12 39 0 1*5 7.48 -6.23

HoHc 17.00 18.50 18.73 18.54 IB 06 15 42 13 71 13 .76 15 26 17 90 18.45 17.65 % Error -17.24 -12 .80 -9.15 0.65 0 78 0 52 5 77 1 .25 5 68 0 96 0.00 -24.30 Wenchi H 17-30 19.06 19.13 19.22 18 66 16 47 Ik 59 14 .75 15 55 19 92 16.91 16.53 % Error -9.70 -6.36 0.67 0.41 3.1*2 2 37 4.64 -2 .50 _e59 -19 .71 -7.26 -10.72

northern .04 20 19.26 Tamale Hc 19.98 21.56 21.58 21.1*2 21.63 19 .84 17.89 16.28 17.92 21 .98 % Error -1.37 -2.53 -19.8 -3.73 -6.60 2 .22 6.20 6.60 -ii* .2 -1 .01 -6.28 -18.38 Yendi H 19.97 21.21 21.90 21.43 21.63 19.52 17.33 16.78 18 .09 21 .04 21.27 19-80 % Error -2.1*1 -1*00 -1.20 2.68 1 .99 6.15 8.84 2.61 2.16 2.05 -1.38 10.06 20 Bole H 20.1*6 21.25 21.27 21.09 21.11 19.16 16.81 15 .91* 16.15 20.31 .19 19.05 % Error -4.l*lt 1.76 -4.83 1.59 1 .10 1.89 3.56 0.25 6.05 -0.50 1 .22 -10.31 . — i. *»• *

TABLE 6

Comparison of measured and estimated H by Eq. (£) for three tovns

STATION JAN FEE MAK APR MAY JUH JUL AUG SEP OCT HOV DEC

Accra H 16.77 19.23 20.52 21.31 20.86 15.91* 16.92 17.17 16.65 20.93 20.32 16.72

17.29 20.76 21.1*8 21.77 21.50 ll*.58 lit. 58 15>9 17.78 21.18 22.1*2 16.55

% Error -3.10 -8.00 -l*.7O -2.20 -3.10 6.50 13.80 9.80 U.70 1.20 10.30 1.00

Kumasi H 15. ll* 16.76 17.92 18.1*7 18.31 15-68 13.38 12.71 ll*.12 15.57 17.61* ll*.29

13.62 15.Ik 21.11 22.92 23.05 17.80 11.98 11.88 17.28 16.15 18.16 13.62 Ic % Error 10.00 -11.80 -17.80 -2l*.10 -25.90 -13.50 -10.50 6.50 -22.1*0 -3.70 -3.10 l*.7O

Takoradi H 15.92 18.57 20.15 20. Ut 18.22 lit. 21 15.1*0 15.62 17.61 lS.66 20.09 16.20

H 16.1*8 20.23 21.27 21.90 18.15 I1*. 37 15.35 15.60 lit. 1*9 18.61 21.90 17.17 c % Error -3.50 -8.90 -5.50 7.10 0.1*0 1.10 0.33 0.13 17.70 0.56 -9.00 -6.00

TABLE 7

Estimation of daily diffuse solar irradiation by five correlation methods. Eqs. (6)-(l0)

STATION JAN FEB MAR APR MAY JUN JUL AUG SEP OCT UOV DEC

Accra H. 7.30 7.50 7.80 7.56 7.19 7.83 7.93 8.05 7.90 7.21 6.5I* 7.08 dg 6.19 6.31 6.53 7.25 6.88 6.39 6.62 6.71 6.73 6.89 5.97 6.0lt X 6.71 6.92 6.00 8.71* 9.18 10. oi* 10.66 10.30 9.70 8.16 6.30 7.19 \ 8.26 8.86 9.9k 10.66 10.93 10.73 11.39 11.15 IO.9I4 10.11* 8.57 6.63 _ y 6.95 7.1*8 7.93 8.03 7.79 7.1*9 7.66 7.70 7.73 7.61* 7.13 6.80

Kvimasi H 7.27 7.86 8.20 8.03 7.76 7.88 7.61* 7.68 8,06 7.83 7.27 7.19 s 6.13 6.61* 6.91* 6.81* 6.63 6.65 6.53 6.67 6.90 6.60 6.26 6.06 \ 7.1*2 7.21 7.71 7.S* 6.1*2 9.M 9.77 9.15 8.05 8.72 7.76 6,86 8.53 8.65 9.25 9-53 9.88 10.19 10.06 9.1*6 8.81* 9.63 9.2k 7.9I* —\ y 6.70 7.32 7.78 7.81* 7.70 7.1*9 7.06 7.00 7.31 7.21 7.0l* 6.52

Tamale H. 5.90 6.61 e.oit 7.82 7.68 7.1*5 7.83 8.17 8.07 6.71 6.13 6.71 d6 5.1*0 6.01* 6.79 6.92 6.80 6.5Q 6.7I* 6.83 6.78 6.12 5.63 5.78 \ lt.73 5.55 5.76 7.1*3 6.70 8.52 10.1*1* 11.50 8.63 6.01* 3.75 l*.O7

7.23 8.02 7.71* 9.52 8.87 10.31 11.58 12.15 9.58 8.1*6 6.1*6 6.10 % 6.62 7.20 7.68 8.08 8.0l* 7.95 7.96 7.85 7.50 7.31* 6.76 6.1*1 FIGURE CAPTIOUS

Fig. 1 Variation of monthly average daily total solar radiation.

Fig. 2 Variation of monthly average daily total solar radiation.

Fig. 3 Variation of monthly average daily total solar radiation.

Fig. k Variation of mean monthly total solar radiation for coastal region. d d

ON \D Fig. 5 Variation of mean monthly total solar radiation for central region. CM CO o o o Fig. 6 Variation of mean monthly total solar radiation for northern region.

o o o Fig. 7 Variation of computed monthly average daily diffuse radiation for Accra.

o o o Fig. 8 Variation of computed monthly average daily diffuse radiation for Kumasi.

ON 00 O m in -3- 9 Fig, 9 Variation of computed monthly average daily diffuse radiation for Tamale. d d d

O-i ON \O odd

D CM fO j j f> a o o o

o o o

^t c^ j- ro on rn d o d cy o co ... o o o

pn o ON en -*• c\j odd

t— H [— in J cu o o o •t

HQMEUCLATUBE

a, "o, climatologieally determined regression coefficients,

D the average day of the year given for each month, • Accra Takoradi solar constant 1353 W/n2 [5]. OOO Saltpond Ill Tf \ 1 *\ monthly average daily total radiation received on a horizontal surface 18 *\ \ J / A \\ 1 / / J A 2 1 hr MJm" day" . \ 1 & \\ /// i\ Vl \ v it< \ / calculated monthly average daily total radiation received on a 16 \ -2 -1 \ horizontal surface MJm day , I \ V Js monthly average daily diffuse radiation received on a horizontal i -2 —1 surface MJm day . w 14 monthly mean daily radiation on a horizontal surface in the absence of o any atmosphere at a particular latitude MJm day" .

12 -. -J 1 1 JFMAMJJASOND n monthly average daily number of hours of observed bright sunshine Month hr aay"1. g solar declination.

o characteristic declination, the declination on which the extraterrestrial irradiation is identical to its monthly average value [12], Kumasi + latitude. Ho ui sunset hour angle degrees. OOO Wenchi

JFMAMJJASOND

-15- -16. Monthly Av.Daily Total Radiation (MJrri9)

Mean Monthly Total Solar Radiation MeanMonthly Total Solar Radiation Mean Monthly Total Solar Radiation 10 in o o o a> o o o o o o

i S -

Diffuse Solar Radiation (MJ m~ ) Diffuse Solar Radiation (MJm"*) o> oo o ro

m m rti rnm JQ p n an Diffuse Sola rRadiat ion (Mdm ! oo o ro

D r 1

•t